Professional Analog Mastering.
Professional Analog Mastering.
226K+ subscribers
Mastering is pretty simple - the job is to balance the frequency response in a way that best represents the song, increase the perceived loudness while preserving dynamics, and establish a balance between differing info in the left and right channels and identical info in the left and right channels.
If you can do that well, that’s 90% of making a great sound.
And you don’t need complex tools.
A fully parametric M/S EQ to control the frequency response and images, a saturator and/or an RMS compressor to control dynamics and fill the frequency response, an upward processor to bring up quieter details while retaining peaks and a versatile limiter is all you need.
The goal is to make each filter, each change, each setting effective - this way we don’t have to add more processors than necessary.
The more conflicting processing you have, the more you need to use another processor to undo what you did with the last one, and the more processed and unpleasant the sound becomes.
The chain we’re creating is simple, with each plugin doing its part well. First is an EQ to balance the frequency response before processing.
Next, we’ll control the RMS or average loudness with this free Kotelnikov compressor. This reduces the need for peak compression later on.
Upward processing and saturation come next. For a cleaner master, just the upward processor will work since it adds mild saturation.
If you want more distortion, then introduce the saturator, and then use the upward processor.
The inflator is great for bringing up quieter details, but each one has a unique character. The knee function of vintage warmer 2 is a personal favorite. The MM-1 maximizer is a great option as well.
With the dynamics controlled, and quieter details brought up, we’ll control the tonal balance with broad EQ filters.
Then, the limiter will introduce the last bit of loudness while controlling peaks in a way that doesn’t change the timbre.
I’ll show each one of these steps in more detail, but for now, let’s listen to the before and after with each processor enabled one at a time.
Before we introduce any processing that could amplify more of something we don’t want, let’s balance the frequency response.
When mastering, a linear phase EQ is a good idea. If kept to a low latency setting, the pre-ringing distortion is very minimal and affects transients less than a zero latency EQ.
With it, attenuate the side image lows with a high pass filter. Since we’re using linear phase we can use an 18dB or 24dB per octave filter without messing up the phase rotation.
I recommend increasing this filter to somewhere between the fundamental range and the 2nd-order harmonic range. This will make everything below the filter identical in the left and right channels.
A small dip in the low mids around 250Hz will reduce masking to higher frequencies and reduce any muddiness present in the mix.
Then, I’ll find some resonances in the high mids - the vocal’s nasal resonances and other instruments will typically make this area too busy, so a small dip helps.
Of course, this will vary between songs, but these filters are almost always useful when balancing a track before subsequent processors.
When compressing a master, the goal is to not be able to hear it. I want to achieve about a dB to 2 of attenuation, and a mix of peak and RMS.
This kotelnikov compressor is free and lets you blend peak and RMS compression. I’ll set it to almost complete RMS compression, but then control a small amount of the peaks as well.
With these settings, you shouldn’t be able to hear compression what so ever, other than that it’s lower in volume.
By controlling the peaks and RMS so subtly, the limiter we use at the end won’t have to attenuate more than a few dB.
This will keep the limiter from sounding too noticeable - so it’s really more of a practical insert we’re using at this point.
Let’s take a quick listen to the EQ and RMS compressor being introduced. Notice that the first EQ helps balance the response, and the compressor doesn’t make much of an audible change.
Watch the video to learn more >
Introducing upward processing is kind of a game-changer. It’s often skipped, but when combined with saturation and limiting, it helps bring up detail and create a loud sound, without the need for heavy limiting.
Saturation reshapes the peaks. Limiting ideally retains the peak shape entirely and uses clean gain to attenuate them when needed.
Upward processing reshapes the wave from the quieter points up, which typically retains the timbre, and the impact of transients, while again, increasing loudness.
Each upward processor shapes a little differently, which means they introduce unique harmonics. Let’s listen to the Oxford inflator and the vintage warmer 2 using the knee function, and we’ll see which one complements the track better.
Watch the video to learn more >
This is the fun part of mastering in my opinion. This is when the dynamics are balanced, details are up, and now we can add everything we want more of.
Again, I’ll set the EQ to a low-latency linear phase setting.
Also, I’ll use another side-image high-pass filter. You might notice that the processing we used introduced info back into the side image of the lows, so it helps to clean that up at this point.
As I said before, it varies from song to song, but a dip in the low mids if needed, then a boost to the lows and the high mids helps a lot.
In the mid-range, a side-image bell makes the stereo image more impressive without creating an unbalanced frequency response.
The air frequencies can be boosted on the side alone if you want to expand them, in the mid if you want to focus them, or identically to the mid and side.
What you choose is up to you, but I find that this is a great starting point more often than not.
Lastly, it’s time to limit the signal.
Whichever limiter you have is fine, but I love this Elephant Limiter by Voxengo.
It lets me control just about every parameter - from the attack to release shape, the time, if the signal expanded into the limiter, if the channels are linked, and so on.
Regardless of the limiter, I’d recommend a release time above 100ms, channel linking around 60-80&, and only 3dB of attenuation at most.
If you have to attenuate more than that to get to the LUFS you want, then it’s best to use your other processors to make the signal louder and more controlled before limiting it.
For example, some additional upward processing combined with a little more peak attenuation will help you achieve a louder, more controlled sound.
Measure the loudness with a LUFS meter, but most importantly, reassess the sound. Does it still sound the way you want when quieter details are pushed forward with the limiter?
If not, check the tone-shaping EQ - this is where you can easily reshape the sound and adjust the balance.
Let’s take a listen to this last EQ and the limiter being introduced.